Association of Early Interventions With Birth Outcomes and Child Linear Growth in Low-Income and Middle-Income Countries

Key Points Question Which interventions under the domains of nutrition, deworming, maternal education, and water, sanitation, and hygiene can improve birth and linear growth outcomes during the first 1000 days of life in low-income and middle-income countries (LMICs)? Findings This study used Bayesian network meta-analyses of 169 randomized clinical trials including 302 061 participants and showed that several nutritional interventions demonstrating greater associations with improved outcomes compared with standard of care, while other domains generally did not. Interventions provided to pregnant women generally demonstrated greater associations with improved outcomes than interventions provided to infants and children at later periods. Meaning The study findings suggest that it is important to intervene early for child development in LMICs, during pregnancy if possible, and combine interventions from multiple domains and test for their effectiveness.


Review ID
The effect of balanced protein energy supplementation in undernourished pregnant women and child physical growth in low-and middle-income countries: a systematic review and meta-analysis Balanced protein energy supplements 7 RCTs, quasi-RCTs, and observational study Buppasiri 2015 5 Calcium supplementation (other than for preventing or treating hypertension) for improving pregnancy and infant outcomes The impact of antibiotics on growth in children in low and middle income countries: systematic review and metaanalysis of randomised controlled trials  Tajikistan or  Tadzhikistan or Tadjikistan or Tadzhik or Tanzania or Thailand or Togo or  Togolese Republic or Tonga or Trinidad or Tobago or Tunisia or Turkey or Turkmenistan or Turkmen or Uganda or Ukraine or Uruguay or USSR or Soviet Union or Union of Soviet Socialist Republics or Uzbekistan or Uzbek or Vanuatu or New Hebrides or Venezuela or Vietnam or Viet Nam or West Bank or Yemen or Yugoslavia or Zambia or Zimbabwe or Rhodesia).hw,ti,ab,cp. 9 ((developing or less* developed or under developed or underdeveloped or middle income or low* income or underserved or under served or deprived or poor*) adj (countr* or nation? or population? or world)).ti,ab. 4060 10 ((developing or less* developed or under developed or underdeveloped or middle income or low* income) adj (economy or economies)).ti,ab. 15 11 (low* adj (gdp or gnp or gross domestic or gross national)).ti,ab. 39 12 (low adj3 middle adj3 countr*).ti,ab. 528 13 (lmic or lmics or third world or lami countr* (Vitamin* or provitamin* or mineral* or trace element* or provitamin* or vitamin*A or vitamin*B or vitamin*B1 or vitamin*B2 or vitamin*B6 or vitamin*B12 or niacin or vitamin*C or vitamin*D or vitamin*E or folic acid or iron or IFA or zinc or copper or selenium or iodine or calcium or MMN or multi*vitamin* or multiple micronutrient or magnesium (Vitamin* or provitamin* or mineral* or trace element* or provitamin* or vitamin*A or vitamin*B or vitamin*B1 or vitamin*B2 or vitamin*B6 or vitamin*B12 or niacin or vitamin*C or vitamin*D or vitamin*E or folic acid or iron or IFA or zinc or copper or selenium or iodine or calcium or MMN or multi*vitamin* or multiple micronutrient or magnesium

No. Terms Hits Comments
South Africa or Sudan or Suriname or Surinam or Swaziland or Syria or Tajikistan or Tadzhikistan or Tadjikistan or Tadzhik or Tanzania or  Thailand or Togo or Togolese Republic or Tonga or Trinidad or  Tobago or Tunisia or Turkey or Turkmenistan or Turkmen  (Vitamin* or provitamin* or mineral* or trace element* or provitamin* or vitamin*A or vitamin*B or vitamin*B1 or vitamin*B2 or vitamin*B6 or vitamin*B12 or niacin or vitamin*C or vitamin*D or vitamin*E or folic acid or iron or IFA or zinc or copper or selenium or iodine or calcium or MMN or multi*vitamin* or multiple micronutrient or magnesium (Infant or infan* or newborn* or new-born* or baby or babies or neonat* or perinat* or postnat* or child* or kid* or toddler* or youth* or pediatrics or pediatric* or paediatric* or peadiatric*).mp. 163817 4 or/1-3 163817 Population terms combined 5 Developing Countries/ 702 LMIC terms 6 (Africa or Asia or Caribbean or West Indies or South America or Latin America or Central America).hw,ti,ab,cp.
7824 Tadjikistan or Tadzhik or Tanzania or Thailand or Togo or Togolese  Republic or Tonga or Trinidad or Tobago or Tunisia or Turkey or  Turkmenistan or Turkmen or Uganda or Ukraine or Uruguay or USSR or  Soviet Union or Union of Soviet Socialist Republics or Uzbekistan or Uzbek or Vanuatu or New Hebrides or Venezuela or Vietnam or Viet Nam or West Bank or Yemen or Yugoslavia or Zambia or Zimbabwe or Rhodesia).hw,ti,ab,cp. 8 ((developing or less* developed or under developed or underdeveloped or middle income or low* income or underserved or under served or deprived or poor*) adj (countr* or nation? or population? or world)).ti,ab. 4060 9 ((developing or less* developed or under developed or underdeveloped or middle income or low* income) adj (economy or economies)).ti,ab. 15 10 (low* adj (gdp or gnp or gross domestic or gross national)).ti,ab. 39 11 (low adj3 middle adj3 countr*).ti,ab. 528 12 (lmic or lmics or third world or lami countr* (Vitamin* or provitamin* or mineral* or trace element* or provitamin* or vitamin*A or vitamin*B or vitamin*B1 or vitamin*B2 or vitamin*B6 or vitamin*B12 or niacin or vitamin*C or vitamin*D or vitamin*E or folic acid or iron or IFA or zinc or copper or selenium or iodine or calcium or MMN or multi*vitamin* or multiple micronutrient or magnesium).mp. (fortified food or food fortified or fortified, food or dietary product* or dietary food* or specialized food or foods, specialized or dietary supplement* or supplement, dietary or supplements, dietary or food supplement* or supplement, food or supplements, food or lipid based supplement or lipid supplement or soy based supplement or soy supplement or fatty acid supplement or omega fatty acid supplement or DHA or EPA or fish oil or long chain fatty acid (Maternal education or mother education or education, maternal or education, mother or breast feeding promotion or breastfeeding promotion or breast feeding education or breastfeeding education (Vitamin* or provitamin* or mineral* or trace element* or provitamin* or vitamin*A or vitamin*B or vitamin*B1 or vitamin*B2 or vitamin*B6 or vitamin*B12 or niacin or vitamin*C or vitamin*D or vitamin*E or folic acid or iron or IFA or zinc or copper or selenium or iodine or calcium or MMN or multi*vitamin* or multiple micronutrient or magnesium).mp. 1901525 18 exp Dietary Supplements/ 5340

No. Terms Hits Comments
Food supplements 19 (fortified food or food fortified or fortified, food or dietary product* or dietary food* or specialized food or foods, specialized or dietary supplement* or supplement, dietary or supplements, dietary or food supplement* or supplement, food or supplements, food or lipid based supplement or lipid supplement or soy based supplement or soy supplement or fatty acid supplement or omega fatty acid supplement or DHA or EPA or fish oil or long chain fatty acid No. Terms Hits Comments promotion or breast feeding education or breastfeeding education).mp. 26 exp Sanitation/ 396291 WASH 27 (waste water management or drinking water or sanitation or sewage disposal or septic tank or latrine* or toilet* or hygiene or WASH or chlorine tablet* or hand washing or water store or potty or soap or detergent (Infant or infan* or newborn* or new-born* or baby or babies or neonat* or perinat* or postnat* or child* or kid* or toddler* or youth* or pediatrics or pediatric* or paediatric* or peadiatric*).mp.  Tajikistan or Tadzhikistan or Tadjikistan or Tadzhik or Tanzania or  Thailand or Togo or Togolese Republic or Tonga or Trinidad or  Tobago or Tunisia or Turkey or Turkmenistan or Turkmen  ((developing or less* developed or under developed or underdeveloped or middle income or low* income or underserved or under served or deprived or poor*) adj (countr* or nation? or population? or world)).ti,ab. 81508 9 ((developing or less* developed or under developed or underdeveloped or middle income or low* income) adj (economy or economies)).ti,ab. 422 10 (low* adj (gdp or gnp or gross domestic or gross national)).ti,ab. 213 11 (lmic or lmics or third world or lami countr*).ti,ab. 5320 12 transitional countr*.ti,ab. 142 13 or/5-12 3416285 LMIC terms combined 14 4 and 13 631731 Population and LMIC terms combined 15 exp Micronutrients/ 604217 Micronutrient supplements and calcium 16 (Vitamin* or provitamin* or mineral* or trace element* or provitamin* or vitamin*A or vitamin*B or vitamin*B1 or vitamin*B2 or vitamin*B6 or vitamin*B12 or niacin or vitamin*C or vitamin*D or vitamin*E or folic acid or iron or IFA or zinc or copper or selenium or iodine or calcium or MMN or multi*vitamin* or multiple micronutrient or magnesium No. Terms Hits Comments promotion or breast feeding education or breastfeeding education).mp. 25 exp Sanitation/ 81650 WASH 26 (waste water management or drinking water or sanitation or sewage disposal or septic tank or latrine* or toilet* or hygiene or WASH or chlorine tablet* or hand washing or water store or potty or soap or detergent Adu-Afarwuah 2015 70 Lipid-based nutrient supplement increases the birth size of infants of primiparous women in Ghana Adu-Afarwuah 2015

NCT00970866
Adu-Afarwuah 2016 71 Small-quantity, lipid-based nutrient supplements provided to women during pregnancy and 6 mo postpartum and to their infants from 6 mo of age increase the mean attained length of 18-mo-old children in semi-urban Ghana: A randomized controlled trial Adu-Afarwuah 2015

NCT00970866
Prado 2016    Nahidi 2011 Effect of early skin-to-skin contact of mother and newborn on mother's satisfaction. Other   Adding multiple micronutrient powders to a homestead food production programme yields marginally significant benefit on anaemia reduction among young children in nepal Population Osei A 2013 Using homestead food production program as a platform to deliver multiple micronutrient powders to infants and young children in Nepal Other   Lactational amenorrhea is associated with child age at the time of introduction of complementary food: a prospective cohort study in rural Senegal,West Africa.

Outcome
Singh H 2017 Daily supplementation with 400 iu vitamin d in term breast fed infants from 0-6 months and changes in total and bone specific alkaline phosphatase-a rct Outcome  Maternal and child supplementation with lipid-based nutrient supplements, but not child supplementation alone, decreases self-reported household food insecurity in some settings    Kramer 2007 Effects of prolonged and exclusive breastfeeding on child height, weight, adiposity, and blood pressure at age 6.5 y: evidence from a large randomized trial

eAppendix. Details to our feasibility assessment and statistical analyses Feasibility assessment of network meta-analyses
Feasibility assessment for network meta-analysis was done after systematic literature reviews for each of the life periods were completed.
We first compared the trial characteristics of included studies in terms of unit of randomization (individual versus cluster-based), blinding, geographic locations (i.e. continents and countries), organized by and across the intervention domains. We also compared the participant characteristics of all included trials. For the pregnancy NMA, we compared the participant characteristics in terms of trimester and mean gestational age at randomization, and follow-up and treatment duration. For the NMA for the exclusive breastfeeding period, we considered mothers' age at enrolment and gestational age at birth, and sex in terms of proportion of boys. For the complementary feeding NMA, we considered age of children and sex also in terms of proportion of boys recruited. The risk of bias was assessed using Cochrane risk of bias assessment tool in all of the included trials across the three life periods, without specific associations for the pre-specified outcomes.
Networks of evidence were then assessed for all of the outcomes in order to assess the connectivity of all included trials. The trials that could not be connected into the network were excluded from the analyses.

General conversions for continuous change from baseline data
For continuous outcomes, the effect measure was the mean difference between the change from baseline (CFB) of one treatment arm versus another treatment arm. Thus, the data input to the statistical model were the mean change from baseline for each arm and their associated standard errors. When trials do not report CFB, this can still be calculated for a specific time point by subtracting the endpoint value from the value at baseline. In this case, the associated standard error can only be approximated by assuming some correlation between the endpoint and baseline value. As a base case, we assumed a correlation of 0.8, but tested scenarios where 0.5 and 0.9 are assumed. The approximated standard error of the CFB was obtained via the basic formula: SE(CFB) = �Var(Endpoint) + Var(Baseline) -2 * Corr(Endpoint, Baseline) * SE(Endpoint) * SE(Baseline) Where Var(X) denotes the variance of X, SE(X) the standard error of X, and Corr(X,Y) the correlation between X and Y.

Statistical models and effect measures
The treatments of interest were compared using Bayesian indirect comparisons following the Bayesian models recommended in the NICE TSD2. 1 Effect measures employed were relative risk for binary outcomes and CFB for continuous outcomes. To this end, in the Bayesian framework logistic regression were used for binary outcomes and linear regression were used for continuous outcomes. All approaches were employed under the random-effects framework. For the Bayesian models, the random-effects models recommended by NICE TSD2 were employed, however, due to the limited number of studies per comparison, a random-effects model with an empirically informed prior on the heterogeneity parameter were also employed. [2][3][4] Model fit was assessed by comparing the Deviance Information Criterion (DIC) of the fixed-effect and random-effects models, as well as their respective leverage plots. All Bayesian models were performed in R using the R2WinBUGS package. 5,6 Three separate chains and sufficiently many iterations (as indicated by Markov Chain Monte Carlo diagnostics) were used to confirm proper burn-in, and an additional number of iterations will be used to construct posterior distributions for all parameters. Longer chains were used should convergence diagnostics be deemed unsatisfactory. The median and 2.5 th and 97.5 th percentiles from the posterior distributions constituted the best estimate of effect and 95% credible interval for each of the comparative effects and other relevant parameters (e.g., the between-study heterogeneity).

Inclusion of cluster trials
When including data from a cluster randomized trial with estimated intraclaster correlation (ICC) and average cluster size in a network meta-analysis that also includes (non-cluster) randomized clinical trials, we make the following adjustments, as recommended by Uhlmann et al 2017. 7 • the design effect = 1 + ( − 1) is evaluated • The effective sample size is calculated. The effective sample size of an intervention arm of size is = / .
• Similarly, when the outcome of interest is a dichotomous one, the number of cases is also adjusted, namely: = / .
• Standard errors of mean differences are now adjusted to the effective sample size, that is: = � .

Primary outcomes
As a first round of indirect comparisons, the above models and effect measures were utilized to establish inference of comparative efficacy and safety of the interventions included in the networks.

Meta-regressions
When there are enough studies for a given network, it may be possible to perform a meta-regression analysis where the relative treatment effect of each study is a function of not only a treatment comparison of that study but also an effect modifier. In other words, with a meta-regression model we estimated the pooled relative treatment effect for a certain comparison based on the available studies, adjusted for differences in the level of the effect modifier between studies. Meta-regression analysis can help explain between-study heterogeneity and minimize (bias) in indirect comparisons due to transitivity violations. We may then model: for fixed effects, where reflects the impact of study level covariate . The random effect modification is as usual, = 0 Other examples where meta-regression may be useful include incorporating baseline risk (as a continuous covariate) in multiple treatment comparison with binary outcomes, or equivalently, a mean baseline continuous outcome.

Assessment of inconsistency
Where closed loops exist in an evidence network, it is important to check that the results from the direct evidence are consistent with the results of the indirect evidence (i.e., that consistency assumption is not violated). To the extent the consistency assumption is violated, results from a network meta-analysis may not be valid. Our assessment of inconsistency contained the following approaches: First, data were be modelled using an independent means model, which contrary to the convention NMA model does not impose the assumption of consistency. 8 The DIC model fit statistics and comparative effect estimates of the two models were compared. If the independent means model appeared to provide a better fit, there was evidence of meaningful inconsistency. Secondly, the methods commonly known as edge-splitting were employed. 8 With this approach, a full network less a single pair wise comparison was modelled and results of the full NMA were compared to the direct evidence of the single pair wise comparison. This step was repeated until such a comparison has been made between every single pair wise comparison involved in at least one closed loop. In other words, the edge-splitting approach allowed for each pair wise direct evidence to be compared to the full body of indirect evidence.
For networks where closed loops exist, the above inconsistency tests were employed (where applicable). If evidence of inconsistency was identified, further statistical analyses were planned to explore the sources of inconsistency or increase consistency in the network (e.g., by meta-regression adjustments or subgroup analyses).

Network meta-analysis models to synthesize direct and indirect evidence
Fixed and random effect meta-analysis At the center of the NMA lies a generalized linear model with a likelihood and a link function that are chosen to reflect the nature of the data at hand. The likelihood was defined in terms of an unknown parameter γ , for treatment in study (e.g., a proportion of success within a binomial process). A link function g(•) was used to transform the parameter on a scale of all real numbers (such as a log of the odds or logit in the case of a proportion): θ = g�γ �. The transformed parameter was in turn modelled using a model that takes the following form: where µ represents this (transformed) outcome in trial j with comparator treatment A. d is the underlying treatment effect of B versus A on a normal scale that is the same for each study, . With the randomeffects meta-analysis model, δ is the trial-specific relative treatment effect of B relative to A. These trialspecific relative effects were drawn from a random effects distribution: δ~( , σ 2 ): The basic NMA model When the available evidence consisted of a network of multiple pairwise comparisons (i.e. AB-trials, ACtrials, BC-trials, etc.) the standard fixed effects model for NMA can be specified as follows: